“FAST MAPPING” FROM HIGH RESOLUTION SATELLITE IMAGES: A
SUSTAINABLE APPROACH TO PROVIDE MAPS FOR DEVELOPING COUNTRIES
M. Gianinetto a, A. Giussani b, G.M. Lechi a, M. Scaioni b
a
b
Politecnico di Milano, Dept. I.I.A.R., Remote Sensing Lab., P.zza L. da Vinci 32, 20133 Milano, Italy
Politecnico di Milano – Polo Regionale di Lecco, Dept. I.I.A.R., via M. d’Oggiono 18/a, 23900 Lecco, Italy
e-mail: {marco.gianinetto, alberto.giussani, giovanmaria.lechi, marco.scaioni}@polimi.it
KEY WORDS: Developing countries, QuickBird, SPOT, IKONOS, EROS, High Resolution, Mapping, Sustainable Development
ABSTRACT:
Imagery coming from high resolution sensors seems to become in the near future a tools to derive maps, comprehending large scales
as well. At the current state-of-the-art, this dream is still infeasible: the main reason is the unavailability of stereo-pairs (barring few
exceptions). Results of different researches on this topic has stated the usefulness of this kind of data to yield cartography in those
countries where production based on traditional methods cannot be really afforded. This impossibility is due to the high skilfulness
of operators involved in the process, to the expensive hardware required (stereo-plotters or DPW), to the availability of a company
which may provide aerial photos over the interested area.
On the contrary, the use of satellite imagery is simpler: data can be purchased via Internet, choosing between different kinds of
sensors and resolutions, processing can be completely performed by commercial SWs, control points can be measured by GPS,
operations are simple and can be easily standardized. Furthermore, the technology transfer process from researchers to operators is
sustainable.
Obviously the cartographic product that might be obtained in this way is different with respect to a numerical 3D map derived from
aerial photogrammetry. The third dimension of object cannot be computed, due to the unavalability of stereo-pairs, and the pixel size
does not allow to derive maps larger than scales up to 1:10,000-1:5,000. Due to difference from the traditional cartography, either in
the production process and in the content of maps, we have termed such a product “Fast Mapping”.
1. BACKGROUND
One of the topic challenge which have engaged the Earth
Observation community in the recent years can be summarized
by the following question: which is the practical use of high
resolution satellite imagery (HRSI)? In spite of the large volume
of research outputs yielded on this subject, the use of HRSIderived products is relatively low.
According to some addresses given by several studies carried
out at national and over-national levels (Holland et al., 2002;
Devriendt et al., 2003; Holland & Marshall, 2003), the spatial
resolution of this kind of data allows to reach the level of detail
to resolve individual objects in the landscape, in a similar way
than the airborne data does. As consequence, HRSI might be
used as source for extraction of detailed, object-related
information and for the production of large-scale maps. The
pixel size of IKONOS images is corresponding to the
information content of aerial images with scale 1:80,000, that
of QuickBird to aerial photos at scale 1:50,000, resulting in the
possibility of deriving topographic maps up to a scale in the
range
1:6,000÷1:10,000.
Concerning
orthophotomaps,
IKONOS images could be used up to scale 1:8,000 and
QuickBird up to 1:5,000 (Jacobsen & Passini, 2003).
In traditionally well-mapped countries, such as West Europe
and North America, where high-quality maps already exist,
aerial photogrammetry is generally available and the
infrastructure to process airborne imagery is well established,
then the actual cost of high resolution space data is preserving
its use in operational environments. The high interest in
carrying on researches on this field is however motivated by the
fact that the price of HRSI will lower in the years to come, as
more suppliers of this kind of data will enter the market.
On the contrary, in countries where experience in mapping and
aerial photogrammetry is not developed, and furthermore the
infrastructure required to collect and process them is not
available, the use of HRSI to derive topographic maps and
orthophotos is an excellent opportunity at the present as well. In
this context, satellite imagery can provide a rapid and highquality data source for the production of different kinds of
maps, such as vector maps, orthophotos, thematic maps and the
like.
In this paper an approach to derive a such kind of cartography is
discussed, giving some addresses to set up an operational
process termed as “fast mapping” (Caprioli & Tarantino, 2001).
As illustrated in the sequel, the proposed issue is not restricted
to an operational concern. Recently, many papers have been
reported to tackle typical problems about HRSI: orientation
(Büyüksalih et al., 2003; Fraser et al., 2002a, 2002b, Fraser &
Yamakawa, 2003, Tao & Hu, 2001, 2002, Valadan Zoej &
Sadeghian, 2003), orthophoto production (Jacobsen & Passini,
2003), image classification (Neubert & Meinel, 2003). On the
other hand, analysis of which features can be effectively and
correctly derived from high resolution data is still to be further
considered. Important results have been published by the
Ordnance Survey of UK, which have tried to extract different
kinds of features from IKONOS images (see Holland et al.,
2002, Holland & Marshall, 2003).
In order to better explain the topic issues of “fast mapping”
(FMAPP) production process, an example of geocoded vector
layers extracted form an IKONOS image will be presented
along the paper.
2. FAST MAPPING GUIDELINES
The goal fo FMAPP is to provide mid-scale regional maps to
countries of the so called “developing world”. According to this
purpose, the selection of source data, techniques for
measurement of ground control points (GCPs), and processing
softwares will follow accordingly.
Before decribing some practical production of a such kind of
maps (from par. 3 on), basic guidelines are shown, collecting
them in 5 main aspects.
2.1 Sustainable cost
It would seem an evidence that, speaking about of a cartography
for the developing countries, the economic problem may
represents a real bottle-neck. Results of a research project
leaded by EuroSDR (Holland et al., 2002) have shown a
comparison between costs of upgrading topographic maps at
1:10,000 scale by means of three different data sources (we
limit our interest only to products among those analysed, being
the others of scarce interest for practical use):
1. aerial photos;
2. IKONOS Geo Ortho-Kit;
3. IKONOS Carterra Geo-Product ortho-rectified by PCI
software.
Considering all tasks needed to upgrade maps, such as flight
planning, imagery, aerial triangulation, GCP measurement,
DEM, ortho-rectification and data capture/feature extraction,
the process based on airborne data is the cheapest. The methods
based on IKONOS data (panchromatic images at 1 m ground
resolution are assumed) have results in a 56% of larger costs by
using data-set (2), and only 6% larger in case (3). Authors
noticed that costs of case (1) derived from a particularly
favourable condition, because refers to an important mapping
agency were aerial photography, GCPs, DEM and digital
photogrammetric workstations are already available. The use of
HRSI would be even more actractive in developing countries,
where these resources are not so plentiful, resulting in the only
sustainable approach to provide mapping.
2.2 Fast production process
As stated in its definition, one of the fundamental peculiarity of
FMAPP is the fast acquisition process, due to the need to
provide maps for large regions as well. In order to do this, the
production workflow is the following:
1.
2.
3.
4.
satellite data acquisition (recovering of archives data or
commission of a new capture);
GCPs measurement;
orientation and orthorectification (possibly the DEM
generation, if not already provided);
data capture and feature extraction.
Concerning timely of this process, the critical stage is the data
acquisition. Archives collecting already available images are
directly accessible on-line via WEB by:
Space Imaging (www.spaceimaging.com), which delivers
IKONOS data;
DigitalGlobe (www.digitalglobe.com), which delivers
QuickBird data
SpotImage (www.sirius.spotimage.com), which delivers
SPOT-5 data.
Currently, no archive is available for Eros-A1 data, being
necessary to ask vendors (www.imagesatintl.com) about which
images have been already collected over the interested area.
However, in case images covering large areas are needed as in
case of regional mapping projects, recent archives data will be
very difficult to be found, and images must be ordered. The
time nedeed to schedule the data capture over the required area
may be even of a few months, and images with a too wide cloud
coverage may easily happen. Nevertheless, the problem of
waiting for the data acquisition exist also in case of using aerial
photos.
All the other tasks need a smaller time to be completed,
depending on the mapping organization and not merely on the
image vendor. Furthermore, barring GCP measurement, other
stages are only data processing operations, involving no
logisthic and organizing problems.
2.3 Reduce need of infrastructures
The principal idea of FMAPP is the use on HRSI, avoiding the
need of aerial photography and of all infrastructures connetted
to this. The only HW and SW requirements are listed in the
following:
•
•
•
•
•
•
workstation for different stages of data processing;
GPS receivers for GCP measurement (either one
master and more rover stations);
GPS data processing SW;
SW for image registration/orthorectification;
mapping SW for data capturecapture/feature
extraction;
GIS SW for management of the resulting spatial
database and for the generation of digital and
hardcopy maps.
2.4 Map contents
The basic geometric map data of FMAPP are digital
orthophotos at mid-scale. It seems that for a not yet mapped
region, a coverage of orthophotos at 1:10,000 scale may be a
very important results for land planning and management. The
availability of ortho-rectified data would allow to reduce the
number of vector information to capture, with the obvious
decrease of time and costs. We retain that, however, some
vector layers should be derived, according to the particular
needs of the country developing the mapping project.
Moreover, information about geographic names should be
externally provided. Modern GPS technology have provided the
users with a large variety of GIS datalogger palm receiver,
which are able to acquire geocoded information directly on the
field, by filling in a pre-defined DB. GPS signal recorded by
these receivers may be processed in a differential mode,
resulting in even sub-meters accuracies (depending on the
distance from the master station). Thanks to a GPS receiver of
such a kind, information that cannot be colletted from the
imagery can be supplied and integrated in the spatial DB.
Collected vector layers and ortho-images have been thought as
the initial data constituting a spatial DB, which will be then
integrated by adding up further information. Each object in the
database is linked to an attribute table, specifying some
important characteristics of it. The attribute table is made up by
a set of attributes which are common to all possible features.
Then specialized attributes are introduced for particular kinds of
features; e.g., in case of roads, attributes describing the class of
the road, which kinds of vehicle can run on it and the like
should be introduced.
2.5 Tecnology transfer
Thanks to its flexibility and independence from local
infrastructures, the FMAPP process is largely prone to be
transferred to the operator in the country where the project is
carried on. Almost all the stages of the production workflow,
barring GPS measurement, consist on data processing
operations, suitable to be accomplished by means of
commercial SWs. These are usually provided by ease-to-use
GUI, by help and tutorial (on-site or on-line via WEB).
Furthermore, the FMAPP process should include an initial
training stage, where personnel involved can be adequately
formed. Also operations concerning data acquisition by GPS
can be performed by not very expert operators, thank to the ease
on use of currently available GIS datalogger.
3. AN EXAMPLE OF FAST MAPPING
A more detailed description of the process to yield maps based
on the FMAPP approach will be proposed in the sequel by
presenting a pratical example.
3.1 Site selection
Due to the location of Politecnico’s laboratory in Lecco (lake of
Como, Northern Italy), we have preferred to use an image
acquired over this area. This fact results in four main
advantages:
1.
2.
3.
4.
a simplification in GCPs (and check points)
measurement and in data acquisition on the ground by
GPS;
the terrain presenting a very large variety of different
scenarios, so that urban, rural, hilly and mountain
contexts could be imaged in the same data set;
the large availability of other cartographic data to
check the results, such as colour orthophotos
(1:10,000), regional raster maps (1:10,000), vector
map of urban areas (1:2,000); furthermore, new data
acquisition are forthcoming and will be used for other
comparisons;
the availability of a DEM.
3.2 Image data set
A pair of adjacent IKONOS panchromatic images have been
acquired over the interested area (see Figure 1) – Lecco
testfield. These data have been collected in June 2001, and are
stored in the SpaceImaging on-line archive. According to the
purpose of FMAPP approach, the lowest price IKONOS
product has been choosen, i.e. the CARTERRA Geo product, a
rectification of the image to a plane with constant height. More
details about the satellite structure and the delivered data can be
found in Gerlach (2000). In Table 2 some important features of
the images are reported. As can be seen, images present a 1,200
m elevation range and could be considered as a very realistic
test.
Figure 1 – IKONOS images over Lecco test field (from
Space Imaging CARTERRA ONLINE).
Image Sensor
Image Type
Spectral Range
Processing Level
Nominal GSD
Interpolation Method
Datum
Map Projection
IKONOS
Panchromatic
450 - 900 nm
Standard Geometrically Corrected
0.84 m
Cubic Convolution
WGS84
UTM-32N
Table 2 – IKONOS image main features.
3.3 Orthoimage generation
The orthorectification process converts imagery into map-like
form by accurately removing all camera and terrain related
distortions. In order to georeferenced images acquired with
spaceborne sensors, two different approaches have been
developed, based on a parametric (physical) and a nonparametric (generalized) model.
Parametric models, based on the collinearity equations, are
physical models that describe the physical imaging process.
They need the knowledge of the sensor model and the position
and the attitude of the sensor during the acquisition.
Non-parametric models are generalized sensor models
(platform independent) that use general functions to compute
the transformation between the image and the ground reference
systems. As mapping function, the Rational Function Model
(RFM), based on the ratios of polynomials with different
degree, is widely used (Tao & Hu, 2001; 2002).
The sensor model for IKONOS images, as like as for the others
HR satellites, is not available to users, but SpaceImaging is
distributing the relation of the Geo-Image to the national
coordinate system in form of rational functions coefficients
(RPCs) - see Tao & Hu (2001, 2002). They do describe the
scene position (rn, cn) as the relation of a polynomial (RFM) as
function of 3D ground coordinates (Xn, Yn, Zn):
rn =
cn =
p1 (X n , Yn , Z n )
p 2 (X n , Yn , Z n )
(1)
p 3 (X n , Yn , Z n )
p 4 (X n , Yn , Z n )
where:
p i (X n , Yn , Z n ) =
m1 m 2 m3
i = 0 j= 0 k = 0
i j k
a ijk X Y Z
(2)
are polynomial functions of maximum power limited to 3 and
aijk are the RFCs’ coefficients.
According to the purpose of FMAPP approach to provide a
mapping coverage at medium scale for developing countries,
where maps are aged or do not exist at all, we have chosen to
orthorectify the IKONOS image with the RFM approach
implemented in commercial SWs using the RFCs supplied with
the IKONOS imagery and a DEM of the area with a step
resolution of 50 m x50 m (see Figure 3).
Recently tests performed by the authors about QuickBird
orthoimage production using non-parametric models, showed
that using sequential polynomial geometric transformations,
such as the RFM and the affine transformation, it is possible to
obtain orthoimages with planimetric precision of about 1.0-1.5
meters using the RPCs supplied with image data and a very few
GCPs (5-6 in our tests).
possible long distances involved (for radio modem) or to the the
lack or the weakness of GSM signal; on the knowledge of the
authors, tests about using satellite mobile phones did not
yielded good results up today.
The second application of GPS in FMAPP is devoted to the
acquisition of vector and GIS information to integrate the
spatial DB. This task can be easily carried out by a GIS
datalogger palm receiver, which allows to collect georeferenced
features (points, lines, polygons) and to fill in their attribute
tables directly on the field. Different classes of these kind of
receiver exist, the most evoluted registering also phase
measurement. This fact result in the possibility of signal postprocessing, by differentiating it with respect to that acquired by
a master station. Accuracy in the sub-meter order for kinematic
points may be reached as far a distance of 30-40 km from the
master. Moreover, some receivers (e.g. Trimble Geoexplorer CE
XT) permits to determine also static points by registering
several epoques during the stationement on the same position.
This possibility, together with the an accuracy under 0.5 m,
might lead to the use of only GIS datalogger as rover receiver,
finalized to both purposes of GCP measurement and GIS data
collection. Practical exploration of this chance in a testfield
would be very interesting for development in FMAPP.
3.5 Cartographic reference system
Figure 3 – DEM of Lecco test site used for orthoimage
generation (step resolution of 50 m x50 m).
3.4 GPS data acquisition
GPS measurement concern two different purposes, i.e.
acquisition of GCPs for HRSI georeference and collection of
GIS data. Both tecniques must be used in relative mode to reach
a sub-meter accuracy To do this, a master station consisting in a
double frequency geodetic receiver must be setup. Moreover,
position of the master station could be computed with respect to
permanent GPS stations, in order to determine its coordinate in
a global geodetic reference system (in practise a realization of
ITRF).
Concerning measurement of GCPs, a rover GPS is moved over
all ground control points to be measured. As illustrated in par.
3.3, HRSI can be used to generate mid-scale orthophotos based
on a very small set of GCPs (usually 10-15 points per image,
covering an area of about 200 km2). The accuracy of their
measurement should be better than the accuracy of image
coordinate. Considering the ground pixel size of about 1 m for
IKONOS, usually GCPs are manually measured so that a
subpixel accuracy is very difficult to be reached. This fact leads
to an accuracy for GPS measurement in the order of σxy=±40-50
cm, which can be obtnained with ease also by using L1 GPS
receiver on long baselines with respect to the master station.
Kinematic and RTK techniques, which could be succesfully
exploited in developed countries, are not yet suitable to be
applied for the developing world. Firstly, the need for real time
measurement does not exist; secondly online communication of
differential corrections may represent a limitation, due to
The reference system of FMAPP must be linked to the use of
GPS measurement for georeferencing satellite imagery,
resulting in a given ITRF realization. This means that the
ellipsoid to be used is WGS84.
The cartographic coordinate system which is the most suitable
is undoubtely the UTM system, based on WGS84 ellipsoid
(UTM-WGS84). This selection would unify cartographic
reference systems used in different countries, following a trend
which in European countries does as well (e.g. in Italy).
Concerning transformation of points into own coordinate
systems of each country, an approximate mathematical relation
and one or more sets of suitable parameters should be provided.
However, this problem is not typical of FMAPP only, but
involves every mapping and surveying activity; in each context
this concerns should be analyzed in detail.
3.6 Vector information capture
The orthoimage derived from HRSI is a potential source of a
very huge information to be extracted. Nevertheless, barring the
work of Holland et al. (2002) and Holland & Marshall (2003), a
still insufficient effort has been carried out so far in analysing
which kind of vector data can be derived.
We performed a small test on this topic, considering a portion
of the orthophotomap generated from IKONOS image over the
area of Lecco, and trying to extract vector layers typical of a
1:10,000 map. In particular, layer representing roads and
buildings have been drawn.
In Figure 4 is reported a pacth from IKONOS image used for
this test, showing an area typical of a mid size town suburbs,
with cottages and intercity roads. In Figure 5 the same
orthoimage with superimposed some extracted vector layers
(buildings in red and roads in yellow) has been depicted.
Extraction of roads sounds to be the easier task, being these
objects well-identified in the image background. More difficult
has resulted the drawing of buildings, especially in case they are
very close to each other or are partially covered by vegetations;
however, in the second case the same problem would sussist
also in case aerial photogrammetry is applied. The availbility of
colour information would largely help in plotting, because
would allow to locate with more ease objects featuring a
contrasting colour, such as roofs, trees, green area, water
surfaces. A realistic possibility would be to use HRSI pansharpened images. Pan-sharpened images are generated by
merging the colour information contained in the lower
resolution visible (or visible/infrared) multispectral bands with
the geometrical information contained in the higher resolution
panchromatic band. The result of the processing is a natural (or
false color) pan-sharpened image, with the resolution of the
panchromatic spectral band.
It should be noticed that reported vector data directly come
from interpretation of the image and have not been edited. In an
operational production process, also this stage would be carried
out, involving recovering of orthogonality, parallelism an the
like.
In the test area, other two cartographic product were available,
i.e. a colour orthophotomap at 10,000 scale and a portion of the
regional raster map (CTR) at the same scale. Because these
maps are geocoded into the Italian official cartographic system
(Datum “Roma40” based on the Hayford ellipsoid) using a
modified Gauss projection (termed as “Gauss-Boaga”), a
transformation to UTM-WGS84 system used for the IKONOS
orthoimage has been computed. Due to the limitation of the
area, a 2D conformal transformation would result enough
accurate.
Vector layers extracted from IKONOS orthoimage have been
overlayed to both orthophotomap (Figure 6) and raster map
(Figure 7). As can be easily understood, roads have been
extracted with a high accuracy, completely conformal to
tollerances adopted for 1:10,000 maps. In Italy, we usually
adopt a cartographic planimetric tolerance which is twofold the
size of the map resolution; in case of 1:10,000 maps, this
tolerance adds up to 4 m.
Of high accuracy and completness is the drawing of building as
well, in particular when compared to the orthophoto.
Comparison with respect to raster map shows several
differences in geometric positions of buildings, but this fact is
probably not due to the quality of the IKONOS orthoimage.
Figure 4 – Patch of the orthoimage derived from an IKONOS
over tha area of Lecco used in the test for evaluting the
possibility for information capture from HRSI.
Figure 5 – Overlapping of the extracted vector layers to the
IKONOS orthoimage patch.
Figure 6 – Orthophotomap (1:10,000 scale) derived from aerial
photogrammetry with superimposed vector layers extracted
from IKONOS orthoimage.
Figure 7 – Raster regional map (CTR at 1:10,000 scale) with
superimposed vector layers extracted from IKONOS
orthoimage.
4. OPEN PROBLEMS AND FUTURE STEPS
In the paper a sustainable approch to provide maps to
developing country referred to as “Fast Mapping” has been
proposed. The image data source are HRSI which are going to
become more accessible in the near future, thanks to a possible
reduction in price, acquisition and delivering time, ease in its
processing.
In order to check possible critical topics and drawbacks in the
production process, an experimental application involving an
IKONOS image acquired over the area of university laboratory
in Northern Italy has been carried out. Findings of this test have
shown that, once the image is delivered, geocoding and
orthoimage generation (if a DEM is available) can be easily
performed by means of commercial software packages and by
using a small set of GCPs. The quality of derived product seems
to be good, either in geometric accuracy, and in the
interpretability of imagery, allowing to extract vector layers. On
the other hand, in case of mapping for a developing country, the
avilability of a DEM with enough accuracy and resolution is the
real bottle-neck. Derivation of DEM from InSAR techniques
would be an interesting solution to this problem.
ACNOWLEDGEMENTS
This work has been carried out under a research (COFIN 2001)
funded by the Italian Ministry for University and Scientific
Research (MIUR), with the title “L’uso delle immagini
satellitari ad alta risoluzione per le analisi territoriali”. We thank
Planetek Italia for the support in IKONOS images purchasing,
and the Lecco Laboratory of Politecnico di Milano for the
support in GPS data acquisition.
REFERENCES
Büyüksalih, G., Kocac, M.G., Oruc, M., Akcin, H., and K.
Jacobsen, 2003. Handling of IKONOS-Images from Orientation
up to DEM Generation. In Proc. of ISPRS-EARSeL Joint
Work. on “High Resolution Mapping from Space”, Hannover,
on CDROM.
Caprioli, M., and E. Tarantino, 2001. Utilizzo del laserscanning per la produzione di cartografia “speditiva”. Bollettino
SIFET, n.2.
Carrion, D., Gianinetto, M., Colombo, A., and. M. Scaioni,
2001. Multi-spectral and multi-temporal imagery registration by
image matching algorithms. International Workshop on GeoSpatial Knowledge Processing for Natural Resource
Management, Varese.
Devriendt, D., Goossens, R., Dewulf, A., and M. Binard, 2003.
Improving Spatial Information Extraction for Local and
Regional Authorities using Very-High-Resolution Data –
Geometrical Aspects. In Proc. of ISPRS-EARSeL Joint Work.
on “High Resolution Mapping from Space”, Hannover, on
CDROM.
Fraser. C.S., Baltsavias, E., and A. Grün, 2002a. Processing of
IKONOS Imagery for submeter 3D Positioning and Building
Extraction. ISPRS Journal of Phot. & R.S., 56, pp. 177-194.
Fraser. C.S., Hanley, H.B., and T. Yamakawa, 2002b. 3D
Position Accuracy of IKONOS Imagery. Photogrammetric
Record, 17(99), pp. 465-479.
Fraser. C.S., and T. Yamakawa, 2003. Applicability of the
Affine Model for IKONOS Image Orientation over
Mountainous Terrain. In Proc. of ISPRS-EARSeL Joint Work.
on “High Resolution Mapping from Space”, Hannover, on
CDROM.
Gerlach, F., 2000. Characteristics of Space Imaging’s Onemeter Resolution Satellite Imagery Products. Int. Arch. of the
Phot., R.S. and Spatial Information Sciences, Vol. 33, Part B1,
pp. 128-135.
Gianinetto, M., Scaioni, M., and A. Giussani, 2003. Cartografia
speditiva da immagini satellitari ad alta risoluzione: un
contributo alla ricostruzione dei paesi colpiti da eventi bellici.
VII Conferenza Nazionale ASITA, Verona, on CDROM.
Holland, D., Guilford, R., and K. Murray (ed.s), 2002.
Topographic Mapping from High Resolution Space Sensors.
OEEPE Official Pubblication, n. 44.
Holland, D., and P. Marshall, 2003. Using High-Resolution
Satellite Imagery in a Well-Mapped Country. Terrain. In Proc.
of ISPRS-EARSeL Joint Work. on “High Resolution Mapping
from Space”, Hannover, on CDROM.
Jacobsen, K., and R. Passini, 2003. Accuracy of Digital
Orthophotos from High Resolution Space Imagery. In Proc. of
ISPRS-EARSeL Joint Work. on “High Resolution Mapping
from Space”, Hannover, on CDROM.
Neubert, M., and G. Meinel, 2003. Evaluation of Segmentation
Programs for High Resolution Remore Sensing Applications. In
Proc. of ISPRS-EARSeL Joint Work. on “High Resolution
Mapping from Space”, Hannover, on CDROM.
Scaioni, M. and M. Gianinetto, 2003. Fusion of aerial and
satellite imagery over the city of Venezia. 2nd GRSS/ISPRS
Joint Workshop on Remote Sensing and Data Fusion over
Urban Areas (Urban 2003), Berlin, on CDROM.
Tao, C.V and Y. Hu, 2001. A Comprehensive Study of the
Rational Function Model for Photogrammetric Processing.
PE&RS, 67(12), pp. 1347-1357.
Tao, C.V and Y. Hu, 2002. 3D Reconstruction Methods based
on the Rational Function Model. PE&RS, 68(7), pp. 705-714.
Valadan Zoej, M.J., and S. Sadeghian, 2003. Rigorous and
Non-Rigorous Photogrammetric Processing of IKONOS Geo
Image. In Proc. of ISPRS-EARSeL Joint Work. on “High
Resolution Mapping from Space”, Hannover, on CDROM.